Monday, August 18, 2014

Irrational exuberance meets secular stagnation

Robert Shiller warns us in the New York Times about the potential risks of high stock market valuations in the US. According to Shiller "the United States stock market looks very expensive right now". Brad DeLong and Dean Baker disagree with Shiller and argue that stock prices might look higher than historical averages but this could be ok given other changes in the economic environment.

Here is a restatement of their debate (and I will be repeating arguments I have made before):

Shiller's concern comes from the fact that price-to-earning (PE) ratios in the US are high by historical standards. Using his own measure, they stand at above 25 which is much higher than the 15 level that was common before the massive 2000 bubble that took that ratio all the way to 44. There is no disagreement about this fact.

Where Brad DeLong and Dean Baker disagree is in how relevant history is an indicator of what constitutes the right level for the PE ratio. The easiest way to think about the PE ratio is to first look at its inverse: earnings as a ratio to prices (EP). This gives us a sense on the yield that a share delivers today (let's keep aside for the sake of simplicity the difference between earnings and dividends). But this is just the static return that shares promise. We know that earnings will be higher in the future and the total return will then be the sum of the "static" return (EP) plus the expected growth of earnings.

Using similar numbers to Dean Baker's post, we can see that historically, a PE of about 15 corresponded to an EP of 6.7%. If we add to this real growth of earnings that was not far from the real growth in GDP, so about 3%, we ended up with a historical return of about 9.7% in real terms. How good was 9.7%? It was a good return. In fact, it was so good that many academics were puzzled by how high this level was. The way they expressed this puzzle is by comparing this return to the returns of alternative assets. Bonds paid somewhere below 4% (real) during most of those decades, which means that stocks yielded close to 6% premium over safe assets (this is what economists called the risk premium and because of its high level it led to a very prolific literature trying to explain the "equity premium puzzle").

Today, a PE of 25 translates into an EP of only 4%. This looks low compared to the historical average and that's where Shiller's concerns come from. But there are three potential reasons why the historical average might not be relevant:

1. Returns on other assets have gone down. Here is where secular stagnation meets irrational exuberance. A new eBook released by VoxEU presents convincing evidence in favor of the idea of secular stagnation. Secular stagnation is characterized, among other things, by real interest rates that are very low, possibly negative. But if this is true, it means that an EP of only 4%, which is about 2.7% lower than the historical one can be entirely justified by equilibrium real interest rates (returns) that are significantly lower than historical ones. What used to be a typical real return for bonds of 3-4% is now 0-1% (or even lower). This means that the return of stocks still remains significantly higher than those of other assets by a margin which is not far from historical levels. In other words, what used to be an EP of 6.7% should now an EP of 4%.

2. Not so fast! Secular stagnation also suggests that the growth rate of real GDP (and therefore earnings) is slowing down. This means that while the static EP ratio might look ok, when we factor in the lower growth of GDP we are back to stocks being too expensive. Correct but what matters here is the relative change in interest rates relative to the growth rate of GDP (earnings to be more precise). CBO projections of potential GDP for the US are lower than historical averages by somewhere around 0.5%. In contrast, interest rates are much lower, on the range of 2-4% relative to historical averages. Let's put these two numbers together: let's assume GDP growth slows down from 3% to 2.5%. This means that with an EP of 4%, stocks promise a return of about 6.5%. Much lower than in the past but still about 6% higher than bonds (if we take 0.5% as the real return on bonds). So we have the same risk premium as in the past and stocks are still a great investment even if prices are high relative to earnings. Of course, if you pick a much more pessimistic growth rate and a higher real equilibrium rate then stocks become expensive again (and if you go in the other direction then they become an even better deal).

3. Finally, why should the stock market risk premium be 6%? Is this a reasonable number? It is certainly be a function of the perceived risk of investing in stocks and arguments can be made in both directions. It is indeed very difficult to argue in favor of a particular number here but a historical perspective is useful: we cannot forget the fact that in the past such a high risk premium led to the conclusion that stock markets were undervalued and that investors were too-risk averse.

One final reflection: as much as I am offering support for the reading that Brad DeLong and Dean Baker do of the today's US stock market, we cannot forget that what will really matter is the perception of the market about how all these variables are likely to evolve going forward. And as Shiller argues, "sociology and social psychology" can be more useful in determining the future returns of the stock market than PE calculations or statements on equilibrium interest rates. But one still hopes that in the long run, fundamental analysis will produce a good indicators of rates of returns. But until the long run arrives, hold tight. As Brad DeLong puts it "If you are not in stocks for the long term, your stock portfolio should not consist of money that you cannot afford to lose."

Antonio Fatás